Modern Machines: Patronage, Information, and Incumbency in Local Elections

Abstract

That incumbents are highly likely to win reelection is a well established pattern at all levels of government in the United States. Scholars have debated the degree to which incumbents are advantaged as a result of selection, responsiveness to voters, or institutional insulation. Research has suffered from endogeneity at the state and federal levels, and at the local level little research exists. This paper begins to address these problems. First, I lay out a theoretical framework for distinguishing between systems in which responsiveness is encouraged versus systems in which incumbents can be reelected without regard to their effectiveness as representatives. I refer to this as the contestability of the political arena. Contestability is measured by the degree to which constituents' information about governmental performance or available alternatives is limited, the degree to which the entry, exit, or behavior of candidates or voters is restricted, and the degree to which incumbents' control over government seats is insulated from their vote share. Then I turn to an empirical analysis of the incumbency advantage in municipal elections. I employ a regression discontinuity design to show that selection effects can not wholly explain the incumbency advantage. Then I provide evidence that low-information elections and large patronage workforces increase the proportion of incumbents who run for reelection and the proportion who win. Finally, I show that spending is disconnected from demographic change in cities with low-information environments.